Hiromitsu Shimakawa

Acquisition and service of temporal data for real-time plant monitoring

Data sampled from plants are registered as temporal data, combined with time stamps. We developed a real-time data server which periodically acquires temporal data and transmits them to clients for graphic presentation. The real-time data server consists of multi threads scheduled based on the rate monotonic theory. This paper explains the design of a real-time data server. The prediction for schedulability of it based on the rate monotonic theory has been verified with an experiment on an actual system.<<ETX>>

The Time/Place/Object model for tracking and history management in manufacturing line control

Tracking and production history management for manufacturing lines requires the ability to retrieve traced-back production performance data from raw materials to final products. Requirements are not so simple. At first, manufactured objects are split into or combined with other objects. Since traditional temporal models concentrate on time-varying or time-series objects, the change of the unit of object management cannot be handled. Secondly, not only object branching and merging but also object dividing and accumulating need to be represented. Conventional workflow models focus on only branching and merging, because of the characteristics of the workflow domain. Thirdly, production histories need to be retrieved in different perspectives according to the purpose of the retrieved data. However, the view of the temporal models is focusing on only the time dimension, while the view of workflow history is only a projection from entire history. We propose the Time/Place/Object model to satisfy the requirements in manufacturing line management and describe the structure and the behavior of a manufacturing line. The model provides the definite categories of five process primitives, three object forms, four invocation logics of ECA rules, and three views of production history. We have developed a family of middleware based on the model and applied it to a tracking system for a steel mill plant.

Digital Audio and Video in Industrial Systems

In industrial environments such as power plants, automated factories, sewage treatment facilities, railways, etc., digital audio and video play at least three important roles:, On-line documentation and training. Pre-stored video in documentation databases is typically viewed interactively, both during routine operation and during emergency situations. 9 Monitoring and surveillance. Video cameras posted around plants allow operators to keep track of security and proper operation and to provide a visual record for subsequent auditing and analysis. 9 Sensors for plant control. Video and image processing are being used increasingly in the automated operation of the plant i t se l f for example, in equipment to measure speeds, count objects, search for production flaws, detect wear of machinery, etc. In these kinds of settings, it is often desirable to integrate many different functions into the same network for example, functions or applications with hard real-time requirements, continuous media such as audio and video, functions requiring rapid response, and traditional applications using traditional data protocols such as TCP/IP. At first glance, this may not seem too difficult if one simply dedicates a portion of total network bandwidth to the traffic with hard real-time requirements, then a portion of the remainder to audio and video, etc. However, bandwidth is only one of the resources and problems that must be considered in a complete network system. Because of the widely different communication requirements of these functions, the demands of their traffic characteristics, flow control, constraints, and performance criteria are typically more challenging than they would be in typical local area or office networks with workstations, PCs, client and server machines, etc. In this position paper, we discuss a number of issues regarding industrial networks, digital audio and video in those networks, and implications on current research directions. These are considered in the context of ATM (Asynchronous Transfer Mode) networks having speeds ranging from 100 megabits/second to one gigabit/second.

Real-Time Reactions in Supervisory Control According toData Freshness

Supervisory control systems in plants mustacquire data without any loss and react for state changes. Someof state change should be reacted in a real-time manner, whileothers may be notified in a best-effort manner. To implementa middleware which works as a server in supervisory control systemsin an acceptable cost, we value the data freshness which meanshow much time has passed from the occurrence of the data. Inthis paper, we present the Active Ring model whichincorporates the ECA mechanism into circular areas. With rulesusing only the latest data, the ECA mechanism realizes time-criticalreactions as firm real-time tasks. In addition to that, the rulestrigger best-effort notification which is realized as a softreal-time task. The middleware works in numerous plants for steelmill, water purifying, and tunnel ventilation. This paper includesan experimental result revealing that the ActiveRing model is effective to provide service according to the datafreshness.

Agents to guide operators with recognition of time series

A practical distributed artificial intelligence (AI) system is proposed for the guidance of plant operations in a steel galvanizing plant. The system includes three autonomous agents for guidance. Each agent records data sampled every moment in an appropriate form for a task assigned to it. The data correspond to states of a monitored entity. Each agent recognizes not only a current state but also a state transition in the course of time to select the best moment for a plant operation. An experiment in a real plant has proved that the plant-operation-supporting system can guide plant operations in much the same way as an experienced operator.<<ETX>>

Production information management for batch manufacturing plants based on ECA mechanism and view generation

Hirom|tsu Shim~kAwa IESL, Mitsubishi Electric Corp. R;~ ~i~waOcon.sdl.melco.co.j p Morik~zu TakegAk| IESL, Miteubishi Electric Corp. t Rkegski@con.sdLmelco.co.jp In recent years, high functional systems for production information management are required. For batch mannf*~-tari~ plant management, both physical process data and product data needs to be handled integratedly. In this paper, the architecture and mechanisms to realize a database system for the batch m~-ufacturing plant management are described in three features. First, the database system is constructed in two layers to integrate data acquisition part and high functional part. As the second feature, to trace a batch or lot which passes through some processes, a flexible data mA,. agement model based on the object-oriented data model and the ECA mechAniMm is presented. For the third feature, collected physical process data and product data are integrated and provided for operators in three views. A prototype system is also introduced with an example of a coffee m~k;n 8 plant. 1 Introduction In recent years, high functional systems for production information management are required. Manufacturing companies need to reduce the production cost, improve the yield-rate and use manufacturing lines effectively. For supporting such activities, production information management systems[5] should collect production data reliably and enable to use the data in various aspects such as recipe management or parameter tuning. In this paper, the architecture and mecbaniRms of a production information management system for batch manufacturing plants are described. To realize production information management systems which enable to use batch manufacturing plants efficiently, we point out and focus on the following three problems and requirements. First, the plant control system and information management system should be connected directly without interfering each other. Data acquisition task requires real-time capability for collecting process values without any loss. On the other hand, data utilization task requires high functionality for providing Pennilion to make digital/hard eopi~ of all or part oflhi~ mala.ial for I~emJ e¢ dmroem ,~-is granted wilhout fee provided that the eopia are not made w di~lm~ for profit er ~,~-ial advanu~ the copy. right notke, the title ofthe publicatlon and ~ date appar, and naive is given tlmt cm, pytight w by permim tfftbe ACM. Inc. To copy otherwir, e, to n~mblish, to I~ en u~d:vs or Io vedislribute to lid require apecifi c p~imim and/~ fee DART 96 Rockvdk AID. USA comprehensible information about plant running. Since conventional database systems such as commercial RDBs do not …

TACT-An Environment to Construct Reliable Systems with Planning-based Task Scheduling

This paper proposes TACT, an environment to construct reliable systems where periodic tasks are dominant. TACT provides a planning based task scheduling mechanism, which enables the master machine to duplicate its task plan on the slave machine in run-time. Through the duplication, tasks on the slave machine get equivalent to those on the master machine. TACT allows two machines to stand-by alternately. The paper describes the concept, functions, and implementation of TACT. It also shows application of TACT to a steel mill control system.